Re: [eigen] internal compiler error in cachefriendlyproduct.h |

[ Thread Index |
Date Index
| More lists.tuxfamily.org/eigen Archives
]

Benoit Jacob schrieb:

Great, this is really worth backporting to the branch!

`
`

please wait a little bit:
My message was a false positive.

`1. The msvc in my office and at home is still not satisfied by the
``patch. I really dont know why!
``There are ?arbitrary? circumstances, witch led my compiler hang up, but
``not eon every run !!.
`

`2. Is this still to be called "/* explicit vectorization */"? Please
``give me some basic information about that. Then maybe i can look out for
``an other, smarter, solution.
`

`3. There several places in cachefriendlyproduct.h witch possibly could
``cause the same problem (by arbitrary circumstances?).
``What I'm mean is: Why does line #504 (below "process remaining coeffs
``(or all if there is no explicit vectorization)") or line #522 (below
``"process first unaligned result's coeffs") not causing the exact same
``problem????
`
4. Rev #917843 cause several warnings and errors (see some of them below):
Cheers
Frank

`Warning 1 warning C4348:
``'Eigen::DiagonalMatrixBase<CoeffsVectorType,Derived>::construct_from_expression'
``: redefinition of default parameter : parameter 3
``c:\develop\eigen\eigen\src\Core\DiagonalMatrix.h 53 test_hyperplane
`

`Error 9 error C2248:
``'Eigen::DiagonalMatrix<_Scalar,_Size>::m_coeffs' : cannot access
``protected member declared in class
``'Eigen::DiagonalMatrix<_Scalar,_Size>'
``c:\develop\eigen\eigen\src\Core\DiagonalMatrix.h 63 test_hyperplane
``Error 14 error C2666: 'Eigen::Quaternion<_Scalar>::operator *' : 2
``overloads have similar conversions
``c:\Develop\Eigen\test\geometry.cpp 268 test_geometry
``Error 15 error C2666: 'Eigen::Quaternion<_Scalar>::operator *' : 2
``overloads have similar conversions
``c:\Develop\Eigen\test\geometry.cpp 328 test_geometry
`